I’ve started to keep a collection of readable papers (mostly engineering/experimental science, but a little bit of more theoretical work as well) that cover what I consider to be pretty foundational concepts, because additional citations are ~free and it at least gives an entry point for readers that are on the edge of the target audience.

As a consequence, I’ve read papers that were written anywhere from the late 19th century to just a year or two ago. In my experience, the older papers tend to be more understandable at a conceptual level but more modern ones tend to be more precise with the details. There is likely some survivorship bias here, though, as there has been more time for the worst of the old papers to be forgotten.

The writing style has also changed a lot— The older papers present things in a much more narrative way, and it can be a challenge to bring the kind of motivational context that allows into a paper that would feel at home in a modern publication.